Targeting Carbohydrate Mimetics of Tetrahydrofuran-Containing Acetogenins to Prostate Cancer

The high potency of the tetrahydrofuran-containing acetogenins (THF-ACGs) against a broad range of human cancer cell lines has stimulated interest in structurally simpler mimetics. In this context, we have previously reported THF-ACG mimetics in which the THF and butenolide moieties of a mono-THF-ACG were replaced with carbohydrate and thiophene residues, respectively. In the present study, towards the targeting of these carbohydrate analogues to prostate cancer (PCa), we synthesized prodrugs in which a parent thiophene or butenolide congener was conjugated through a self-immolative linker to 2-[3-(1,3-dicarboxypropyl)ureido] pentanedioic acid (DUPA), a highly specific ligand for prostate-specific membrane antigen (PSMA), which is overexpressed on prostate tumors. Both prodrugs were found to be more active against receptor positive LNCaP than receptor-negative PC-3 cells, with 2.5 and 12 times greater selectivity for the more potent thiophene analog and the less active butenolide congener, respectively. This selectivity for LNCaP over PC-3 contrasted with the behavior of the parent drugs, which showed similar or significantly higher activity for PC-3 compared to LNCaP. These data support the notion that higher activity of these DUPA-derived prodrugs against LNCaP cells is connected to their binding to PSMA and suggest that the conjugation of PSMA ligands to this family of cytotoxic agents may be effective for targeting them to PCa.


Introduction
Natural product-inspired small molecules remain a primary line of cancer therapeutics. In this context, tetrahydrofuran-containing acetogenins (THF-ACGs) have attracted attention for their high potencies against a broad range of human cancer cell lines [1][2][3][4]. Their unusual mechanism of action, which may in part be due to mitochondrial ATP disruption, may make them particularly effective against tumors that are significantly dependent on oxidative phosphorylation [5][6][7][8]. Accordingly, a variety of THF ACG analogues have been examined, with a major focus being on structures in which more complex segments of naturally occurring compounds have been replaced with simpler or easily accessed substitutes [3]. In this vein, we recently reported carbohydrate analogues of naturally occurring ACG 4-deoxyannomontacin (4-DAN) 1, wherein the THF and butenolide moieties were replaced with mannose and thiophene residues, respectively ( Figure 1) [9]. These analogues 2 and 4 were prepared and screened as a mixture of C-10 epimers, as earlier studies on similar carbohydrate congeners suggested that the configuration at this central carbinol position should not impact on cytotoxicity, and for simplification of the synthesis [10]. Similar to 4-DAN, these mimetics showed cytotoxicity in the low-to-submicromolar range against two human prostate cancer (PCa) cell lines, PC-3 and LNCaP. Although their mode of action remains to be determined, 2 and 4 are attractive therapeutic leads given the prevalence of PCa [11,12], their relatively simple syntheses, and the possibility that their sugar motifs may confer favorable tumor selectivity and solubility properties. Against this backdrop, we considered targeting PCa with prodrugs of 2 and 4 in which a tumor vector is conjugated via a self-immolative linker to the C-10 carbinol [13][14][15]. For this preliminary study, we employed an established methodology using 2-[3-(1,3-dicarboxypropyl)ureido] pentanedioic acid (DUPA) as a vector for the prostatespecific membrane antigen (PSMA) and a known disulfide-based linker. PSMA is a transmembrane glycoprotein that is overexpressed on a variety of primary and metastatic PCa and is a target for cancer diagnostics and therapeutics [16][17][18]. DUPA binds PSMA with high specificity and affinity and the receptor promotes endocytosis of bound ligands [19]. DUPA-linked conjugates of imaging and cytotoxic agents have shown significantly increased selectivity for PSMA-expressing tumors [20][21][22][23]. Drug release from prodrugs 3 and 5 is facilitated by interaction with thiols, which exist in higher concentration in cells than in plasma, thereby reducing the possibility of nonspecific uptake of the parent drugs 2 and 4, respectively [16][17][18][19][20][21][22][23][24]. Thus, the guiding hypothesis is that the DUPA ligand would elicit selective uptake of the prodrugs in PSMA-positive cells over PSMA-negative cells, thereby leading to increased activity against the former. Herein, we describe the synthesis and initial cytotoxicity data for 3 and 5.

Synthesis
The synthesis of the DUPA thiol segment started with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDCI)-promoted coupling of the known DUPA carboxylic acid 6 and the HCl salt of amino tosylate 7 [20][21][22][23]25] (Scheme 1). This reaction delivered the chloride 8, which presumably resulted from the substitution of tosylate with in situ chloride. Treatment of 8 with potassium thioacetate yielded 9. Exposure of 9 to trifluoroacetic acid in the presence of triisopropylbutylsilane (TIBS) provided 10 [26]. For prodrug 3, the reaction of 11, an intermediate in the earlier synthesis of 2, and 12, the p-nitrophenyl carbonate of the 2-pyrido disulfide of 2-mercaptoethanol, followed by the removal of the acetal-protecting groups in the product, provided 13 [9,27,28] (Scheme 2). Finally, disulfide exchange on 13 with the in situ-generated thiol derivative of 10 delivered the prodrug conjugate 3. A similar sequence of reactions on 14, the precursor to 4, led to prodrug 5 [9]. Prodrugs 3 and 5 were characterized by NMR and HRMS (Supplementary Material).

Cytotoxicity Data
The cytotoxicity of 2-5 against the two human prostate tumor cell lines PC-3 and LNCaP was determined by the XTT cell viability assay after 48 h of treatment with the test compounds (Table 1 and Supplementary Material). LNCaP is known to overexpress the PSMA receptor, whereas PC-3 is receptor-negative [29]. The thiophene parent drug 2 showed low-micromolar-nanomolar-range activity, with eight times higher activity against PC-3 compared to LNCaP. The butenolide drug 4 was at least an order of magnitude less potent compared to 2, with very similar activity against both cell lines. In contrast, both prodrugs 3 and 5 were more selective for LNCaP than PC-3, albeit at lower potency compared to their respective drugs 2 and 4. Prodrug 3, the derivative of the thiophene drug 2 that was more selective for PC-3 over LNCaP cells, showed a reversal in selectivity, with approximately 2.6 times selectivity for LNCaP. Interestingly, the prodrug 5, the derivative of the less potent butenolide drug 4 that was similarly active against both cell lines, was 12 times more selective for LNCaP over PC-3. In comparison, doxorubicin, which is clinically used against PCa, was found to be 3.2 times more active against LNCaP cells [30]. That both prodrugs 3 and 5 are more selective against LNCaP cells compared to their parent drugs 2 and 4, respectively, which are more or similarly active against PC-3 cells, supports our hypothesis that the selectivity of these prodrugs is promoted by their binding to-and internalization by-PSMA on LNCaP cells. More detailed studies, for example, using cotreatment of cells with prodrug and a competing PSMA ligand, will be needed to confirm whether DUPA recognition is the basis for selectivity. Investigations with stable drug-DUPA conjugates that are not degradable intracellularly, as well as quantification of intracellular drugs and prodrug concentrations, will also be needed to interrogate whether the active drug entity is in fact the original drug. In the latter context, it could be argued that the activity of these prodrugs is primarily due to the DUPA moiety and not the original drug. However, this seems unlikely, given the fact that both 3 and 5 have the identical DUPA residue but are appreciably different in potency.
It is also noteworthy that the LNCaP selectivity observed for prodrug 5 is appreciably higher than for prodrug 3. This higher selectivity may be connected to the lower potency of 4, the parent drug in 5, compared to the drug entity in 3. A contributing factor may be the fact that 4 shows no selectivity for either cell line, whereas 2, the parent drug in 3, appears to be intrinsically selective for PC-3 compared to LNCaP cells. In the context of increasing selectivity of this class of DUPA conjugates to PSMA-expressing cells, known DUPA linkers that decrease overall prodrug hydrophobicity and increase PSMA affinity are envisaged [20][21][22]. Potency can be improved by increasing the activity of the drug entity, which may be possible by modifications in the lipid segments, as observed for the parent class of THF ACGs and analogues thereof [1-3,31,32].

Materials and Methods
Unless otherwise stated, all reactions were carried out under a nitrogen atmosphere in oven-dried glassware using standard syringe and septa technique. NMR spectra were obtained on 500 or 600 MHz instruments. Chemical shifts are relative to the deuterated solvent peak or the tetramethylsilane (TMS) peak at (δ 0.00) and are in parts per million (ppm). Assignments for selected nuclei were determined from 1 H COSY and HSQC experiments. Thin-layer chromatography (TLC) was performed on 0.25 mm thick precoated silica gel or C18-W silica, HF254 aluminum sheets. Chromatograms were observed under UV (short-and long-wavelength) light, and/or were visualized by heating plates that were dipped in a solution of ammonium (VI) molybdate tetrahydrate (12.5 g) and cerium (IV) sulfate tetrahydrate (5.0 g) in 10% aqueous sulfuric acid (500 mL). Flash column chromatography (FCC) was performed using silica gel 60 A (230-400 mesh) or C18 (Carbon 17%) 60 A (230-400 mesh) and employed a stepwise solvent polarity gradient, correlated with TLC mobility. Hexanes used for FCC had a boiling point in the 40-60 • C range.

IC 50 Determination
Serial dilutions of compounds to be tested were made in culture media from 1 nM-1 mM concentration. The determined cell number from the standard graph was plated in a 96-well plate in triplicate per each dilution of the compounds. The cells were allowed to attach overnight by incubation at 37 • C in a CO 2 incubator. The media from the well were removed and media with serial dilutions of the compounds were added to the cells. The cells were incubated with the compound for 48 h at 37 • C in CO 2 incubator. Activated XTT reagent was added to the wells and incubated for 2 h. Absorbance from the wells was measured at 475 nM and background at 660 nM. Background subtraction was made, and all the data were plotted in prism and IC 50 values were determined.

Conclusions
Two prodrugs comprising carbohydrate mimetics of mono-THF-containing acetogenins conjugated via a self-immolative linker to a ligand for PSMA, were prepared. These conjugates were found to be approximately 3 and 12 times more active against LNCaP, a PSMA receptor-positive prostate tumor cell line, compared to PC-3, a PSMA receptor-negative prostate tumor cell line. This result contrasted with the activity of the nonconjugated cytotoxic agents, which showed similar or higher activity against PC-3 compared to LNCaP. These data suggest that targeting PSMA may be viable strategy for modulating the tumor selectivity of this family of cytotoxic agents and new directions for carbohydrate-based therapeutics. Studies on improving prodrug selectivity and potency using modified DUPA linkers and drug entities, as well as assays against a wider panel of human tumor and normal cells are directions for future work. Mechanistic and drug development investigations, including cytotoxicity assays in the presence of competing PSMA ligands, and determination of drug uptake, stability and cellular distribution will also be performed.